feat: support video download for zhihu.com #1203

Author: MisakaTAT

app/register.go

@@ -39,4 +39,5 @@ import (
 	_ "github.com/iawia002/lux/extractors/yinyuetai"
 	_ "github.com/iawia002/lux/extractors/youku"
 	_ "github.com/iawia002/lux/extractors/youtube"
+	_ "github.com/iawia002/lux/extractors/zhihu"
 )

extractors/zhihu/types.go

@@ -0,0 +1,17 @@
+package zhihu
+
+// minimum field
+type video struct {
+	PlayList struct {
+		FHD resolution `json:"FHD"`
+		HD  resolution `json:"HD"`
+		SD  resolution `json:"SD"`
+	} `json:"playlist_v2"`
+}
+
+// minimum field
+type resolution struct {
+	Size    int64  `json:"size"`
+	Format  string `json:"format"`
+	PlayURL string `json:"play_url"`
+}

extractors/zhihu/zhihu.go

@@ -0,0 +1,90 @@
+package zhihu
+
+import (
+	"encoding/json"
+	"fmt"
+	"github.com/iawia002/lux/extractors"
+	"github.com/iawia002/lux/request"
+	"github.com/iawia002/lux/utils"
+	"github.com/pkg/errors"
+	"strings"
+)
+
+const (
+	videoURL = "www.zhihu.com/zvideo"
+	api      = "https://lens.zhihu.com/api/v4/videos/"
+)
+
+func init() {
+	extractors.Register("zhihu", New())
+}
+
+type extractor struct{}
+
+func New() extractors.Extractor {
+	return &extractor{}
+}
+
+func (e *extractor) Extract(url string, option extractors.Options) ([]*extractors.Data, error) {
+	if !strings.Contains(url, videoURL) {
+		return nil, errors.WithStack(extractors.ErrURLParseFailed)
+	}
+
+	var err error
+	html, err := request.Get(url, url, nil)
+	if err != nil {
+		return nil, errors.WithStack(err)
+	}
+
+	videoID := utils.MatchOneOf(html, `"videoId":"(\d+)"`)
+	titleMatch := utils.MatchOneOf(html, `<title.*?>(.*?)</title>`)
+
+	if len(videoID) <= 1 {
+		return nil, errors.New("zhihu video id extract failed")
+	}
+
+	title := "Unknown"
+	if len(titleMatch) > 1 {
+		title = titleMatch[1]
+	}
+
+	resp, err := request.GetByte(fmt.Sprintf("%s%s", api, videoID[1]), url, nil)
+	if err != nil {
+		return nil, errors.WithStack(err)
+	}
+	var data video
+	if err = json.Unmarshal(resp, &data); err != nil {
+		return nil, errors.WithStack(err)
+	}
+
+	streams := make(map[string]*extractors.Stream)
+	resolutions := map[string]resolution{
+		"FHD": data.PlayList.FHD,
+		"HD":  data.PlayList.HD,
+		"SD":  data.PlayList.SD,
+	}
+
+	for k, v := range resolutions {
+		stream := &extractors.Stream{
+			Parts: []*extractors.Part{
+				{
+					URL:  v.PlayURL,
+					Size: v.Size,
+					Ext:  v.Format,
+				},
+			},
+			Size: v.Size,
+		}
+		streams[k] = stream
+	}
+
+	return []*extractors.Data{
+		{
+			Site:    "知乎 zhihu.com",
+			Title:   title,
+			Streams: streams,
+			Type:    extractors.DataTypeVideo,
+			URL:     url,
+		},
+	}, nil
+}

extractors/zhihu/zhihu_test.go

@@ -0,0 +1,29 @@
+package zhihu
+
+import (
+	"github.com/iawia002/lux/extractors"
+	"github.com/iawia002/lux/test"
+	"testing"
+)
+
+func TestDownload(t *testing.T) {
+	tests := []struct {
+		name string
+		args test.Args
+	}{
+		{
+			name: "video test",
+			args: test.Args{
+				URL:   "https://www.zhihu.com/zvideo/1620162752064061440",
+				Title: `Cursor, GPT-4 驱动的强大代码编辑器 - 知乎`,
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			data, err := New().Extract(tt.args.URL, extractors.Options{})
+			test.CheckError(t, err)
+			test.Check(t, tt.args, data[0])
+		})
+	}
+}